Volcanological Survey of Indonesia, MVO, Yogyakarta, Indonesia

Multiplets, i.e. events with similar waveforms, are common features on active volcanoes. The seismograms of multiplets are analyzed by cross-spectrum techniques: this procedure improves by a factor of about 10 the precision of differential P-arrival times and therefore the accuracy of the relative location of earthquakes. Long period events which cannot be located because of the impossibility to pick up P-waves on individual seismograms can be located with a precision of about 10 m. Such a precision permits fault planes to be mapped inside a volcanic edifice and the azimuth and strike of fractures to be defined. Seismograms of the two events (of a doublet) that occur on different dates are analyzed by the Cross Spectrum Moving Window technique (CSMW) for measuring the time delay between waves in the coda. The pattern of the delays in the coda is a function of the temporal changes of seismic velocity that occurred inside the volcano during the time interval that separates the two events of a doublet. We illustrate the potential of the doublet technique for detecting temporal changes inside a volcano by performing computations of synthetic seismograms. The case of a dyke injected inside the volcano is considered as well as that of the replenishment of a superficial magma chamber and of a general increase in velocity in the summit of the volcano. Data from Merapi volcano (Indonesia)illustrate a possible temporal velocity change inside the volcano several months before the 1992 eruption.

One major problem associated with the interpretation of seismic signals on active volcanoes is the lack of knowledge about the internal structure of the volcano. Assuming a 1D or a homogeneous instead of a 3D velocity structure leads to an erroneous localization of seismic events. In order to derive a high resolution 3D velocity model ofMt. Merapi (Java) a seismic tomography experiment using active sources is planned as a part of the MERAPI (Mechanism Evaluation, Risk Assessment and Prediction Improvement) project. During a pre-site survey in August 1996 we tested a seismic source consisting of a 2.5 l airgun shot in water basins that were constructed in different flanks of the volcano. This special source, which in our case can be fired every two minutes, produces a repeatable, identical source signal. Using this source the number of receiver locations is not limited by the number of seismometers. The seismometers can be moved to various receiver locations while the source reproduces the same source signal. Additionally, at each receiver location we are able to record the identical source signal several times so that the disadvantage of the lower energy compared to an explosion source can be reduced by skipping disturbed signals and stacking several recordings.